I think it's pretty clear that during the video shooting, it's mostly shoot-and-stop mode as
@jugulear said, the typical pacing is 2-3 sex scenes (typical 40 minutes each) being successfully shot per a full day of work. I guestimate that it works out to more than 2 full hours of "sex" needed to film enough good footage for 30+ minutes of the final vid. Say the simple but crucial moment of ejaculation, it might take the crew 2-3 tries to get the camera, the lights, everything in the right place from the moment the male actor pull his penis out of the vagina, (off-screen pauses and reposition necessary equipment, test shoot, maybe another pause) shift his body to her upper body, to ejaculate on to her face or chest. So what seems like a continuous sex act, almost-orgasm, quickly pull out, shift body, cum on her face, may in fact be a 5+ minutes procedure.
Of course there are many vids that make it a point to use long continuous shots, either as a mark of authenticity or just plain due to budget concern (shoot a full length vid within 3 hours).
And of course in an orgy vid which is by definition more complicated, the need to micromanagement (and lots of interruptions) is much higher.
One hint that much more real time is needed to shoot a 120 minutes final product is the outdoor genre, too often the critical second half of an outdoor vid happens in during and after sunset causing really lousy lighting. The main idea of outdoor sex is (1) public sex is risque, (2) sex in the sun, both purposes are defeated by darkness, I think the studio and directors know that but it just take that long that the whole vid can't be filmed within the daylight hours (which is pretty long in summer in Japan).
As for the many long running series you mentioned (I want to add just one a bit of my fav: My Girlfriend's Busty Sister/Mother Creampie OK Seduction), well what more you expect to hear? What you see is what you get, everything from dialog and mannerism, costume, specific sex acts and packing, even the cover art is super consistent. Like you said, just drop in each idol's face and name, perfectly predictable. Not just the content of the vid is predictable, it's even predictable which idols will eventually perform in each classic formula series (based on her contract status with studios owning which series).